Research a Product's Real World Before Looking at Other Ads

There's a moment in prep when the references stop being useful. You have a board of thirty frames from other campaigns, and you can name the category's visual grammar from memory: the rake of light across a dark wall, the slow push-in on a hand adjusting a hinge, the wordless shot of someone looking up as if a thought just arrived. You know exactly what the room has already seen. What you don't know is who takes the thing out of the box, whether they read the instructions first, or where it ends up when nobody's watching.

That gap is where the film is. The research that closes it doesn't start with other ads. It starts with the product doing something in a real context — being opened, shared, carried, maintained, worn, stored, worked around — and with the people on either side of that action.

Three piles have to stay separate the whole way through. What you observed (or what your source recorded). What a reference expresses. What you propose to invent. Most weak treatments are weak not because they lack imagination but because those three piles got stirred together, and nobody in the room can tell which claim is carrying weight anymore.

Read the brief for the question it leaves open

A commercial brief almost always contains use information, compressed. It tells you who the product is for, the occasion it belongs to, the mandatory product shots, the approved benefit, and often a line about where the thing is used. Start there. Extract the use, the context and the constraints before you go looking for anything exotic, because those are what your research has to stay answerable to.

Then look for the gap. Briefs describe categories, not moments. A brief can tell you the lamp clamps to a desk without telling you whether the person using it clamps it themselves or stands there while somebody else does it. That single unanswered question is enough to justify a day of looking.

Two things the brief does not authorize, no matter how interesting your research gets. It doesn't let you rewrite an approved benefit, and it doesn't let you quietly swap in a new target audience because the one in the deck is harder to cast. If your observation makes you want to change what the product claims, that's a conversation, not a treatment.

Match the source to the claim you want to make

Different sources can tell you genuinely different things, and they can't be substituted for each other once you're making a factual claim about behavior.

Authorized direct observation records what actually happened, once, with that person, on that day. It is the only kind of source that can support a sentence like "this is how the handover went." It is also the narrowest: one occurrence, one pair of people.

A documented process — the brand's own how-to clip, the assembly insert, a mounting guide — demonstrates the ideal procedure. These sources are excellent on the product and unreliable on the person. A manual is a great witness to how a clamp is supposed to be attached and a terrible witness to whether anyone does it that way at home. Use it to keep your scene physically honest; don't use it as evidence of a habit.

A depiction — another campaign, a stock clip, a friend demonstrating for you — shows an action without recording one. A friend showing you how they'd do it is performing, which is fine and often revealing, as long as you write down that it was a performance.

If you're observing someone in their own space, with their own things, get permission and keep a note of what it covers. It's easier to know what you're allowed to describe later than to reconstruct that after the shoot.

Write down the behavior before you name the mood

Record first, interpret second, and keep them in different columns. What's worth writing down: hand positions, the order of actions, waits, interruptions, what was already occupying a hand, what had to move to make room, how much space the action needed, what the physical constraints of the object forced. Not "she seemed reluctant." Instead: she stopped, looked at the clamp, then continued.

That's the whole discipline. One observed handover can suggest a scene question that will carry a film. It cannot establish typical customer behavior, and it cannot tell you why anyone did anything — those need their own support or their own honesty about being guesses.

A useful test for whether an observation is worth developing: does it involve the product doing something the brief actually cares about; does it leave a question a camera could answer; and can it survive without the reference you're about to attach to it? If the observation only works once you've dressed it in someone else's look, you didn't find a relationship — you found a moodboard entry.

Let the reference interpret, not testify

Here's the three-part separation in practice. When you bring a visual or sound reference into a treatment, it should be doing exactly one job, and you should be able to say which:

1. What was observed. From your source. Stated with its scope.
2. What the reference illustrates. An expressive relationship — how attention is held, how long a shot waits, where the camera refuses to go.
3. What you propose to create. The invented scene, with its invented parts named.

References shift as the shot list gets real, and that's normal. What you can't do is let a reference become evidence about the product's world. Nobody in that reference was using your lamp.

Other advertising has one legitimate research job, and it's a good one: convention radar. If the category keeps returning to the same three moves, that tells you what the client's reviewers have already seen, what will read as familiar, and what you'll have to justify if you go a different way. That's real information about the room you're pitching to. It's simply not information about the product's world — and if it's the only account you have of that world, your treatment is a copy of a copy.

Recognizing a convention is not the same as borrowing its solution. You can note that every ad in the category resolves on a silhouette against a window and then decide, deliberately, to resolve somewhere else — which is a different act from pulling the frame and shooting it.

The lamp handover: one invented note, three separate inputs

The example below is fictional. It's constructed to demonstrate the method, not to report fieldwork, and nothing in it should be read as a finding about how people use task lamps.

The note. A portable task lamp changes hands between two people. The receiver clears one hand. The giver rotates the clamp. The receiver waits through the rotation, then takes the lamp.

That's the entire note, and it's already most of a film. Before you use it, count what it doesn't carry. It doesn't say lamps are typically handed over like this. It doesn't say what was in the receiver's hand before it was cleared. It doesn't say why the giver rotated the clamp — to fit the receiver's table edge, to turn the handle outward, or from a habit the giver couldn't explain. One occurrence, two people, and a motive nobody wrote down.

The reference. A reference that earns its place here isn't a shot of a lamp. It's a piece of observational work whose interest sits in the interval before something changes hands: the camera holds the two-shot through the exchange instead of cutting on it; the sound of the mechanism is the loudest thing in the scene; the frame doesn't lunge toward the object once the object arrives. That's an expressive claim about attention — the moment matters more than the thing. Notice that this description tells you nothing about whether anyone actually waits like that, because it can't.

The scene, proposed. A table at the end of a working day. One person is mid-task, one hand occupied steadying something. The second person arrives with the lamp. The cut lands on the receiver's hand clearing, so we don't watch the lamp enter the frame. The giver checks the clamp and rotates it; we stay on the receiver's empty hand and face while that happens. Then the transfer. Then the light, and the camera stays with the hand as it returns to the task the lamp just joined. The product gets its close-up last, in use, from the position of the work rather than from a turntable.

What's invented in that paragraph: the table, the time of day, what occupies the receiver's hand, the decision to cut on the clearing, the decision to withhold the close-up to the end. What comes from the note: the clearing, the wait, the rotation, the transfer, in that order. If a real observation replaced the invented one, its source, permission and limits would have to arrive with it.

Where the object still binds you. The lamp's real geometry and weight constrain the scene, and finding that out is part of the research. How far the head tilts, what the clamp will grip, how the cable wants to fall — those constraints are what make the proposal shootable rather than notional. A discovered behavior changes where the attention goes, not what the object physically does or what the client approved. If the observation seems to imply something about the product's performance, route that to whoever substantiates claims. It's not a treatment line.

The generic alternative. The beauty version: seamless ground, the lamp rotating on a turntable, a raking beam across a dark wall, a hand clicking the switch, a voiceover about focus. It's competent, it's on-brand, and every sentence it says could be said about any lamp in the category. It answers what does this product look like and nothing about what this object is in someone's day. Which version performs better with an audience is a separate question, and nothing above has tested it. The narrower claim stands on its own: one of these carries information specific to an observed relationship, and the other carries category information dressed as a scene.

End with something traceable

A treatment is easier to defend — in the room, and later on set — when every choice can be traced to where it came from. Line the scene up against its inputs before you send it:

Anything that can't be placed in one of those rows is either an untested assumption or an invention you haven't owned yet. Both are fixable. Both are much harder to fix after the shoot.

The reason to do this before you open the reference boards is not that advertising is a bad source of ideas. It's that the boards will always suggest a finished answer, and a finished answer from someone else's film will keep you from asking the question your product's world is already holding. Find one relationship worth developing. Trace it. Then go find the references that help you shoot that.

Frequently asked questions

Why research the product’s real world before looking at other ads?

References from other campaigns show the category’s visual grammar and always suggest a finished answer. They do not tell you who takes the product out of the box, whether they read the instructions first, or where it ends up when nobody is watching. The research that closes that gap starts with the product doing something in a real context and with the people on either side of that action.

What three piles must stay separate in a treatment?

Keep separate what you observed or what your source recorded, what a reference expresses, and what you propose to invent. Weak treatments are often weak not because they lack imagination but because those piles got stirred together, so nobody can tell which claim is carrying weight.

How do direct observation, documented process, and depiction differ as sources?

Authorized direct observation records what actually happened once, with that person on that day, and is the only kind of source that can support a sentence like “this is how the handover went.” A documented process such as a brand how-to clip, assembly insert, or mounting guide demonstrates the ideal procedure; it can be reliable on the product but not on whether anyone does it that way at home. A depiction, such as another campaign, a stock clip, or a friend demonstrating, shows an action without recording one, and if it is a performance that should be written down.

What is the legitimate research job of other advertising?

Other advertising can serve as convention radar. If the category keeps returning to the same moves, that tells you what the client’s reviewers have already seen, what will read as familiar, and what you will have to justify if you go another way. That is information about the room you are pitching to, not information about the product’s world; recognizing a convention is not the same as borrowing its solution.

What does the traceability table demonstrate?

It lines up each element of the scene with where it came from. For example, one hand clearing before a transfer and the receiver waiting through the rotation can come from the observation note; the camera holding the interval instead of cutting can come from the reference’s expressive relationship; time of day, the table, and the withheld close-up are proposal elements that should be named. Anything that cannot be placed in one of those rows is either an untested assumption or an invention that has not been owned yet.

Further reading

• Write Field Notes That Can Support an Observational-Series Pitch — If your own observation notes need to hold up later, this field-note discipline — dated headers, separated description, interpretation and reflection — transfers directly to recording product behavior.